Multi-Sensor Triangulation of Multi-Source Spatial Data · Multi-Sensor Triangulation (MST) •...

JACIE, March 16, 2006

Multi-Sensor Triangulation of Multi-Source Spatial Data

Ayman Habib, Chang-Jae Kim, and Ki-In BangDigital Photogrammetry Research Group

http://dprg.geomatics.ucalgary.ca

Department of Geomatics EngineeringUniversity of Calgary, Canada

https://ntrs.nasa.gov/search.jsp?R=20070038219 2020-06-15T12:55:59+00:00Z

http://dprg.geomatics.ucalgary.ca/

2JACIE, March 16, 2006

Overview

• Introduction.• Multi-sensor triangulation. • Multi-primitive triangulation:

– Points.– Linear features.– Aerial features.

• Experimental results.• Conclusions and future outlook.


Introduction• There is a tremendous increase in data acquisition

systems, which are available for the mapping community:– Photogrammetric systems:

• High resolution imaging satellites.• Metric analog frame cameras.• Metric digital frame cameras.• Metric digital line cameras.• Medium-format digital frame cameras.

– LIDAR systems.– GPS/INS navigation units.

• These systems provide complementary information.• We need to provide an integrating environment of

these sensors: Multi-Sensor Triangulation (MST).


Photogrammetric SystemsFrame Cameras

Applanix DSS SONY 717

RC10

Kodak 14n Canon EOS 1D

DMC

Line Cameras

IKONOSADS 40


LIDAR Systems

ALS 40 (Leica Geosystems) OPTECH ALTM 3100


Point-Based Triangulation

⎥⎥⎥

⎦

⎤

⎢⎢⎢

⎣

⎡

−−−

=⎥⎥⎥

⎦

⎤

⎢⎢⎢

⎣

⎡

−Δ−−Δ−−

0

0

0

ZZYYXX

Rc

yyyxxx

A

A

AT

pa

pa

λ


Photogrammetry:

Direct measurement of intermediate points on images

Line-Based Triangulation


LIDAR:Plane fitting & intersection

manual identification of LIDAR patches with the aid of imagery



LIDAR:Manipulation of range and intensity images



Image line

LIDAR control line

Projected LIDAR line

Direct incorporation of LIDAR lines as control in the photogrammetric BA

1

2

⎥⎥⎥

⎦

⎤

⎢⎢⎢

⎣

⎡

−−−−−

Κ ′′Φ ′′Ω′′=→

cysdistortionyyxsdistortionxx

RV pi

pi

),,(3

⎥⎥⎥

⎦

⎤

⎢⎢⎢

⎣

⎡

′′−

′′−

′′−=

→

O

O

O

ZZYYXX

V

2

2

2

2

⎥⎥⎥

⎦

⎤

⎢⎢⎢

⎣

⎡

′′−

′′−

′′−=

→

O

O

O

ZZYYXX

V

1

1

1

1

),,( OOO ZYX ′′′′′′),,( OOO ZYX ′′′

),,( Κ′Φ′Ω′

i'i

),,( Κ′′Φ′′Ω′′

( ) 0321 =•× VVVrrr



Patch-Based Triangulation

B(XB,YB,ZB)

LIDAR point

A(XA,YA,ZA)

C(XC,YC,ZC)

P(XL,YL,ZL)

Photogrammetric (A,B,C points) and LIDAR surface patches

Direct incorporation of LIDAR patches as constraints in the photogrammetric BA


Patch-Based Triangulation

0

1111

=−−−−−−−−−

=

ACACAC

ABABAB

AiAiAi

CCC

BBB

AAA

iii

ZZYYXXZZYYXXZZYYXX

ZYXZYXZYXZYX

A

B

C

a b

c

LIDAR point, i

Volume of the pyramid:i, A, B, C should = 0

Direct incorporation of LIDAR patches as constraints in the photogrammetric BA


Multi-Sensor Triangulation (MST)• Developed an integrated triangulation system.

– Multi-sensor: Satellite imagery, aerial imagery, LIDAR and GPS/INS.

– Multi-primitive: distinct points, linear features, and aerial features.

• Advantages:– Takes an advantage of the extended coverage of imaging

satellites.– Takes an advantage of the high geometric resolution of

aerial imaging systems.– Utilizes sparse frame imagery to improve the weak

geometry of imaging satellites while reducing ground control point requirements.

– Uses LIDAR data for photogrammetric geo-referencing.


MST: Experimental Results

IKO

NO

S

DSS Digital

DSS:

Expected accuracy:

planimetric: 0.25mvertical: 0.74mspatial: 0.78m



Upper Block

DSS: Upper Block Upper LIDAR Scan



Middle Block

DSS: Middle Block Middle LIDAR Scan



Lower Block

DSS: Lower Block Lower LIDAR Scan


IKONOS scenes and GCP layout over Daejeon, Korea

Stereo-IKONOS with GCP Layout


Examples of Tie Points



Examples of Tie / Control Lines



. u1B

. u1A

. u1P


Example of a Control Patch


RMSE (m)GPS / LIN / PATCH (mm) Total

0 NONE N/A N/A N/A N/A N/A

PATCH 0.004 3.582 3.394 2.207 5.406

GPS + LIN 0.005 2.116 1.358 1.803 3.093

NONE N/A N/A N/A N/A N/A3.065

3.066

4.190

LIN 0.005 2.105 1.370 1.757

GPS 0.006 2.109 1.048 1.963

GPS + PATCH 0.004 2.374 2.580 2.294

018

X Y Z# GCP # Frames 0σ̂

No Ground Control Points

Scan line direction

138 Control Lines & 139 Control Patches


Scan line direction

Configuration of 5 GCP


5 Ground Control Points

Scan line direction


0 NONE N/A N/A N/A N/A N/A

PATCH 0.004 1.670 1.009 1.759 2.628

GPS + LIN 0.005 1.733 1.112 1.717 2.681

NONE 0.006 1.907 1.084 3.7474 4.3422.716

2.727

2.631

LIN 0.005 1.786 1.115 1.717

GPS 0.007 1.805 1.024 1.770

GPS + PATCH 0.004 1.647 1.008 1.786

518





Scan line direction


0 NONE 0.005 1.442 1.608 3.290 3.936

PATCH 0.004 1.516 1.025 1.774 2.549

GPS + LIN 0.005 1.783 1.132 1.724 2.726

NONE 0.006 1.568 1.029 2.392 3.0392.696

2.618

2.598

LIN 0.005 1.738 1.130 1.722

GPS 0.007 1.657 1.025 1.747

GPS + PATCH 0.004 1.578 1.024 1.792

718





Scan line direction


0 NONE 0.011 1.092 0.870 1.450 2.013

PATCH 0.004 1.110 0.845 1.475 2.030

GPS + LIN 0.006 1.143 0.934 1.498 2.103

NONE 0.008 1.129 0.863 1.528 2.0872.112

2.107

2.026

LIN 0.006 1.173 0.921 1.495

GPS 0.008 1.122 0.859 1.563

GPS + PATCH 0.004 1.095 0.846 1.478

4018




0

5

10

15

20

25

30

35

0 1 2 3 4 5 6 7 8 9 10 15 40

Number of Control Points

RMS

E, m

NO FramesFrameLINPATCHGPSGPS + LINGPS + PATCH

N/A



DSS / IKONOS Ortho-photos



Original Image


Ortho-Photo Generation


True Ortho-Photo Generation


Differential Rectification

Generated Ortho-photo



True Ortho-photo

Generated Ortho-photo



Concluding Remarks

• The introduced methodologies are successful in:– Using LIDAR features for photogrammetric geo-

referencing.• Line-based and patch-based photogrammetric geo-referencing

using control derived from LIDAR data.

– Delivering a geo-referenced imagery of the same quality as point-based geo-referencing procedures.

– Taking advantage of the synergistic characteristics of spatial data acquisition systems.

• The triangulation output can be used for the generation of 3-D perspective views.


Recommendations for Future Work• Automated segmentation of LIDAR data to extract

the patches and linear features.• More investigation into using the outcome from

the geo-referencing procedure for the verification of the system calibration.

• Utilize the raw LIDAR measurements in the patch-based photogrammetric geo-referencing.– Such a utilization will allow for LIDAR system

calibration.• Quality assurance and quality control procedures

for LIDAR data.


3-D Perspective View

Multi-Sensor Triangulation of Multi-Source Spatial Data · Multi-Sensor Triangulation (MST) •...

Documents

Transcript of Multi-Sensor Triangulation of Multi-Source Spatial Data · Multi-Sensor Triangulation (MST) •...